Skip to Main Content

When scientists announced last summer that they had used CRISPR to eliminate a disease-causing gene in human embryos, critics immediately erupted with, “You didn’t!” Some meant it in the sense of, omigod, how could you take this step toward designer babies? Others meant, no, your experiment didn’t do what you claimed it did.

The ethical debate about designer babies will probably last longer than the West Antarctic Ice Sheet, but the scientific brawl about the study’s claims has been mostly resolved. In new papers published Wednesday, two groups of scientists detail why they think the researchers who created the CRISPR’d embryos goofed. But in a reply, Shoukhrat Mitalipov of Oregon Health and Science University and his colleagues reject the criticism, reporting that they re-tested the embryos — and got the same results.


“Our conclusions were right,” Mitalipov said in an interview.

The results, if they stand up to continued scrutiny, reaffirm the original paper’s conclusion that using the genome-editor CRISPR-Cas9 to repair a disease-causing gene in an embryo is far more complicated than expected. While the technique might one day allow couples undergoing in vitro fertilization to have biological children who are spared a calamitous inherited mutation, it’s unlikely to work for introducing entirely new traits — for eye color, say, or strength — into offspring.

For their experiments, Mitalipov and his colleagues started with donated eggs from a woman carrying a healthy version of a gene that, when mutated, causes an incurable heart disorder called hypertrophic cardiomyopathy. Then they used IVF to fertilize the eggs with sperm that carried the disease-causing mutation (but did not let the resulting embryos develop beyond eight cells or transfer them to a surrogate mother).


With one batch of eggs, they simultaneously injected CRISPR-Cas9 molecules that cut out the mutation and supplied repair DNA; with another batch they injected the same CRISPR molecules only after the fertilized eggs, or zygotes, started growing and dividing. The first approach produced a higher percentage of embryos free of the mutant gene (about 73 percent), suggesting that, when it comes to repairing mutations with CRISPR, the sooner after fertilization the better.

The scientists used standard genetic techniques to look for the mutant, paternal gene. They didn’t find it in those 73 percent of embryos, which carried only the healthy gene. Mission accomplished.

Not so fast, the critics said. Recent studies have shown that CRISPR can tear through a genome like a tornado through Kansas, leaving swaths of nothingness. That can make it impossible to tell what CRISPR actually did, as a study last month revealed, which could be why Mitalipov didn’t detect the mutant paternal gene, argued scientists led by Maria Jasin of Memorial Sloan Kettering Cancer Center and, separately, Paul Thomas of Australia’s University of Adelaide. Both critiques and Mitalipov’s response were published in Nature.

“Evidence for correction was not provided in the original article, even as this was the central conclusion,” Dieter Egli of Columbia University, a co-author of Jasin’s critique, said in an email interview. “[But] I’ve come to accept that some of the embryos have been edited.”

Mitalipov knew that CRISPR might delete huge chunks of DNA, he said. After that criticism initially arose, he therefore re-analyzed the embryos’ DNA using a technique that should pick up mega-deletions if that’s why cardiomyopathy-causing DNA seemed to have vanished from the embryos. “We did a very detailed analysis and didn’t detect that,” he said.

Critics also zeroed in on the most surprising, and significant (for designer babies), finding from the CRISPR’d embryos: In every case where CRISPR deleted the paternal, mutant gene as intended, the embryo’s genome repaired it not with the healthy gene that CRISPR carried but by copying the healthy gene on the maternal chromosome. The critics offered several reasons why that couldn’t happen, including that the maternal and paternal genomes were physically too far from each other in just-fertilized eggs.

Again, Mitalipov said he had confirmed this “interhomolog repair,” not only with the original cardiomyopathy gene but with two others. He’s not sure why embryos don’t accept a repair gene, but suspects it may be because the introduced DNA is single-stranded (that just happens to be what CRISPR scientists use), and embryos prefer the natural, double-stranded DNA sitting over on mom’s chromosomes.

Critics were partially mollified: Although “interhomolog recombination is not firmly established, the [new] study does provide more substance than the original,” Egli said.

Coincidentally, scientists at the Massachusetts Institute of Technology recently reported exactly such interhomolog repair in mouse embryos. After something like CRISPR-Cas9 cuts DNA on, say, a maternal chromosome, the cell repairs the break by copying the corresponding part of the paternal chromosome, or vice versa.

If zygotes resist incorporating introduced DNA into their genomes after CRISPR cuts out a mutation, the designs in designer babies will be very limited. Only a gene already on a chromosome from one of the parents could replace a disease-causing or otherwise-unwanted gene.

For DNA not carried by either parent, including for supposedly desirable cosmetic or other traits, “that procedure doesn’t really work,” Mitalipov said. “The external repair template has no chance” against the gene from mom or dad.

That underlines something CRISPR experts agree on: The technique is too unpredictable and unreliable for use in fertility clinics any time soon. “It is uncertain whether gene correction by interhomolog recombination occurred in all of the embryos, some of the embryos, or, in the most extreme case, none of the embryos,” Jasin said. “Given the major consequences of gene editing in human embryos, additional studies are called for.”

Mitalipov’s study, she added, “offers optimism for gene correction in human embryos.”